It’s Probably a Bubble, But There Is Plenty Else to Invest In (Ben Inker)

GMO argues the current AI market resembles the 2000 Internet Bubble more than the 2007 Everything Bubble or 2021 Duration Bubble, and that this distinction matters enormously for portfolio construction. The S&P 500 is trading near its 2000 valuation peak on cyclically adjusted earnings, while signs of speculation include quantum computing stocks trading at 1,007 times sales and AI startups raising capital at $50 billion valuations within months of formation. The key insight is that unlike the 2007 and 2021 bubbles — where every risk asset was overpriced simultaneously — today's bubble is concentrated in U.S. large caps and AI-adjacent names. Non-U.S. equities, deep value stocks, and liquid alternatives are priced to deliver equity-like or better real returns. GMO's risk/reward regression line carries a slope of +0.4, similar to 2000, meaning an equal-weighted portfolio of risk assets offers a much higher expected return than the cap-weighted MSCI World. An agnostic investor can tilt away from AI without accepting a lower long-run expected return. GMO's Benchmark-Free Allocation Strategy currently holds roughly 50% stocks, 28% liquid alternatives, and 20% Treasuries, with a forecast real return of 6.5% versus 0.2% for a 60/40 benchmark. The portfolio rose 17.9% in 2025 through November 12, outpacing both the 60/40 benchmark and the S&P 500, driven largely by non-U.S. and deep value positions rather than AI exposure.